Message-ID: <542EEC9A.3090308@kernel.dk>
Date: Fri, 03 Oct 2014 12:36:10 -0600
From: Jens Axboe <axboe@kernel.dk>
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Thunderbird/31.1.2
MIME-Version: 1.0
To: Benjamin LaHaise <bcrl@kvack.org>
CC: Kent Overstreet <kmo@daterainc.com>, linux-kernel@vger.kernel.org,
        Zach Brown <zab@zabbo.net>, Jeff Moyer <jmoyer@redhat.com>,
        Slava Pestov <sp@daterainc.com>
Subject: Re: [PATCH] aio: Fix return code of io_submit() (RFC)
References: <1412359693-2535-1-git-send-email-kmo@daterainc.com> <542EE753.20005@kernel.dk> <20141003182220.GA17057@kvack.org>
In-Reply-To: <20141003182220.GA17057@kvack.org>
Content-Type: text/plain; charset=windows-1252; format=flowed
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org

On 2014-10-03 12:22, Benjamin LaHaise wrote:
> On Fri, Oct 03, 2014 at 12:13:39PM -0600, Jens Axboe wrote:
>> On 2014-10-03 12:08, Kent Overstreet wrote:
>>> io_submit() could return -EAGAIN on memory allocation failure when it
>>> should
>>> really have been returning -ENOMEM. This could confuse applications (i.e.
>>> fio)
>>> since -EAGAIN means "too many requests outstanding, wait until completions
>>> have
>>> been reaped" and if the application actually was tracking outstanding
>>> completions this wouldn't make a lot of sense.
>>>
>>> NOTE:
>>>
>>> the man page seems to imply that the current behaviour (-EAGAIN on
>>> allocation
>>> failure) has always been the case. I don't think it makes a lot of sense,
>>> but
>>> this should probably be discussed more widely in case applications have
>>> somehow
>>> come to rely on the current behaviour...
>>
>> We can't really feasibly fix this, is my worry. Fio does track the
>> pending requests and does not get into a getevents() forever wait if it
>> gets -EAGAIN on submission. But before the fix, it would loop forever in
>> submission in -EAGAIN.
>
> There are lots of instances in the kernel where out of memory is potentially
> exposed to the user.  If we're failing a memory allocation that is well under
> 1KB, the system is probably completely hosed.
>
>> How are applications supposed to deal with ENOMEM? I think the answer
>> here is that they can't, it would be a fatal condition. AIO must provide
>> isn't own guarantee of progress, with a mempool or similar.
>
> I'm not sure if using a mempool is appropriate for allocations that are
> driven by userland code.  At least with an ENOMEM error, an application
> could free up some of the memory it allocated and possibly recover the
> system.

Since fio just hit this, it has nothing it can potentially free to make 
progress possible. There was no pending IO, so all it can do is quit. 
But I do agree that if a small alloc like that fails, then we are 
probably pretty darn screwed anyway, and it doesn't matter that much 
what we do. My main concern was a potential change in the ABI, but since 
we could already return -ENOMEM from other cases, that is probably a 
moot point.

-- 
Jens Axboe

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/